Modulated Modularity Clustering as an Exploratory Tool for Functional Genomic Inference
Figure 4
MMC analysis of simulated data.
Three measures of accuracy are reported for MMC as the number of observations ranges from 4 to 36 in the simulation study. Shown in red for each case is the percentage of simulations in which pairs of variables are correctly clustered. Shown in blue is the percentage of pairs correctly separated; the overall percentage correct is in black. Each point on the plot represents the results across 1,000 datasets. The scenario considered in Table 1 (MMC default) is highlighted in gray.